RE: ISSUE-67 use of reification in mapping rules is unwise (axiom annotation)

>-----Original Message-----
>From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org]
>On Behalf Of Ian Horrocks
>Sent: Monday, July 14, 2008 12:22 PM
>To: OWL 1.1
>Subject: ISSUE-67 use of reification in mapping rules is unwise (axiom
>annotation)
>
>
>I wonder if we aren't ready to resolve this?
>
>The issue "deals with the rule to do with axiom annotations". We
>already decided to use owl:Axiom in the mapping to RDF. Michael says
>[1] that "I think that this is not really a problem". 
[...]

Oh, please wait! With this statement, I did *not* refer to the question
whether it is a problem to use RDF reification to encode axiom annotations.
Rather, I meant by it that it would probably not be a problem to
additionally allow the custom use of RDF reification in OWL 2 DL - alongside
its use as syntax vocabulary. This is something very different.

Background of my mail was that in OWL 1 DL, the custom use of RDF
reification was actually allowed. The question now was what happens if RDF
reification becomes OWL syntax. I thought that it won't be a problem to
further allow custom usage, i.e. that it would not be a problem to *not*
disallow the custom use of RDF reification in OWL 2 DL ontologies. Actually,
disallowing RDF reification for custom use would be a backwards
compatibility issue.

For the analog question about the custom use of rdf:List, I already stepped
back from my earlier discussion, which went in the same direction as the
mail you cited by [1]. But the technical reasons which brought me to step
back for lists do not necessarily also apply for reification, so I am still
undecided about this question.

Anyway, this all has nothing to do with my stance regarding whether to use
RDF reification for axiom annotations or not. I was always skeptical
("scary" would perhaps be a better notion) about using RDF reification in
OWL 2, mostly for cultural reasons (a lot of people are hostile against RDF
reification). 

I previously proposed to use a shadow vocabulary instead. At that time, this
suggestion did not receive a lot of applause. And yes, it would really be
just a kind of "cultural workaround", leaving most of the technical problems
as they are. However, the backward issue mentioned above would be solved by
this approach, which is really worth thinking about it, I suppose! 

But now that we have decided to use a kind of shadow vocabulary for encoding
negative property assertions (resolution of ISSUE-81), instead of using RDF
reification, I would appreciate to use the same approach for annotating
axioms as well. 

Regards,
Michael

[1] http://lists.w3.org/Archives/Public/public-owl-wg/2008May/0190.html